Profiles2024-2026
Leo Yu Zhang
Griffith University
14Papers
6Last author
32Unique co-authors
0First author
0Solo
First and last author come from byline position, which some venues assign alphabetically.
Recent papers14
- "Do Not Mention This to the User": Detecting and Understanding Malicious Agent Skills in the WildAuthor 7 of 7USENIX Security 2026 · 46 citations
- ARES: Scalable and Practical Gradient Inversion Attack in Federated Learning Through Activation RecoveryAuthor 2 of 7S&P 2026
- Debiased Dual-Invariant Defense for Adversarially Robust Person Re-IdentificationAuthor 7 of 7AAAI 2026
- AdvEDM: Fine-grained Adversarial Attack against VLM-based Embodied AgentsAuthor 10 of 10NeurIPS 2025 · 27 citations
- Detecting and Corrupting Convolution-based Unlearnable ExamplesAuthor 7 of 7AAAI 2025 · 13 citations
- Improving Generalization of Universal Adversarial Perturbation via Dynamic Maximin OptimizationAuthor 4 of 7AAAI 2025 · 7 citations
- Pre-training CLIP against Data Poisoning with Optimal Transport-based Matching and AlignmentAuthor 4 of 8EMNLP 2025 · 1 citation
- BiMark: Unbiased Multilayer Watermarking for Large Language ModelsAuthor 4 of 7ICML 2025
- Not All Edges are Equally Robust: Evaluating the Robustness of Ranking-Based Federated LearningAuthor 3 of 6S&P 2025
- Secure Transfer Learning: Training Clean Model Against Backdoor in Pre-Trained Encoder and Downstream DatasetAuthor 7 of 7S&P 2025
- Robust Backdoor Detection for Deep Learning via Topological Evolution DynamicsAuthor 3 of 8S&P 2024 · 39 citations
- Conditional Backdoor Attack via JPEG CompressionAuthor 5 of 5AAAI 2024 · 21 citations
- Revisiting Gradient Pruning: A Dual Realization for Defending against Gradient AttacksAuthor 4 of 7AAAI 2024 · 10 citations
- Towards Model Extraction Attacks in GAN-Based Image Translation via Domain Shift MitigationAuthor 3 of 7AAAI 2024 · 5 citations
